The objective of stereo matching is, given a point in one view ofa scene, to find the homologous point in another view from an adjacentcamera. We use the discrete cosine transform (DCT) with evolutionarymethods as an approach to solve the correspondence problem. The matchingprocess is implemented by generating for each image block sets of DCTcoefficients, which when reverse transformed become the estimateddisparity map for that block. Assumed statistical properties of the ACcoefficients are taken into account, while the DC component is estimatedfrom a preliminary block matching stage. Matching is performed in theimage domain using an intensity similarity measure. The job ofoptimising the DCT coefficients, whose inverse transform gives thedisparity, is carried out by a biologically inspired optimisationtechnique: the genetic algorithm
展开▼